The main goal of the proposed research is the development of better methods for diagnostic classification of electrocardiograms through efficient use of computer methods. Significant improvement could be demonstrated for an adult male population. A large data base was derived from eight Veterans Administration hospitals. In order to make the computer program applicable to the general population, extension is necessary to include women and children. The latter category particularly requires comprehensive studies of normal and abnormal records since an adequate data base for children electrocardiograms does not exist. It is proposed to collect data from a total of seven children's hospitals in order to expedite this development. Increasing evidence from large epidemiological studies indicates that the electrocardiogram contains significant predictive and prognostic information. Since this has never been systematically investigated, a prospective study of long-term trends was organized with five hospitals participating. Coronary artery disease, hypertensive cardiovascular disease, and chronic obstructive pulmonary disease were chosen for study. Computer analysis of electrocardiograms during and after stress tests suffers from lack of sensitivity and specificity. Through application of multivariate analysis techniques, it is hoped to improve accuracy in recognition of abnormalities.